Effect of oxygen stoichiometry on the critical temperature and thermal

Chem. Mater. 1990,2,421-424

421

Effect of Oxygen Stoichiometry on the Critical Temperature and Thermal Expansion of Two-Layer BiSrCaCu Oxide Superconductors H. M. O'Bryan,* W. W. Rhodes, and P. K. Gallaghert AT&T Bell Laboratories, 600 Mountain Ave., Murray Hill, New Jersey 07974 Received February 22, 1990 Single-phase Bi (Sr,Ca)3Cu20&t6 has been prepared from an initial batching formulation of Bi203,1.5s&03, 1.5CaC03, and 2tuO by solid-state reaction. Oxygen stoichiometry of the compound has been defined by thermogravimetric analysis, and it can be varied by thermal history. Thermal expansions and critical temperatures (T,) have been obtained for different oxygen stoichiometries. A t 500 "C changes of A6 = 0.04, AT, (onset) = -22 K and hL/L = -0.06% are obtained for an oxygen anneal vs a nitrogen anneal. The stoichiometry effects for Bi2(SrCa)3Cu20s+6 are more like Ba2YC~307-6 than Pb2Sr2(YCa)C~308+6. Introduction Since the discovery by Maeda' et al. that the rareearth-element-free BiSrCaCu oxide system yields compounds with attractively high T,'s, much work has been directed at obtaining compositions that yield a single-phase ceramic and have a well-defined T,. It has been found that the kinetics of reaction are slow, and many special schemes have been suggested to increase the rate of formation of the superconducting phase. Among these, a low Po, atmosphere2 and the addition of Pb3t4and other cations have shown success in yielding large amounts of the three-layer (107 K) phase. Obtaining the single phase has been more difficult, and the compositions, which correspond to the simple cation ratios 2212 or 2223, are not single phase even after extended reaction times. The effects of oxygen stoichiometry on T,, lattice parameter, and thermal expansion of the two-layer Pb-free Bi2[SrCa]3C~208-b compound have been reported. Buckley5 found as much as a 16 K shift in T, with increasing quench temperature for samples annealed in air, while Morris6 noted a 13 K increase with decreasing Po, for a 600 "C anneal. This latter shift did not depend on cooling rate. Groen7 used a larger Po, range at 600 "C and obtained a 40 K shift for a com ound with 10% Y substituted for Ca. All reported a 0.1- increase in the c lattice parameter for samples equilibrated at the lower Po,. Buckley calculated an oxygen stoichiometry variation of A6 = 0.4 while the maximum that Groen could measure was A6 0.04. Groen reported an expansion of +0.06% a t 600 "C as the Po, is decreased. A small peak in the thermal expansion near 300 "C was seen only during the initial heating. A phase transition was suggested as the cause of this anomaly. Other workers have not found significant effects of oxygen stoichiometry in the two-layer (-85 K) superconductor.

x

-

Experimental Section Bi203,SrC03,CaC03,and CuO were mixed by ball milling to yield various formulations near the 2212 composition. After mixing, each composition was prereacted in oxygen at 800 "C for 12 h. Zirconia containers were used to hold the powders during prereaction. After a second ball milling with ZrOz media under water, pressed shapes were sintered in oxygen for 100 h at a temperature between 840 and 880 "C. An Ag sheet was used to support the ceramic during firing. X-ray diffraction (XRD)was used to establish whether reaction was complete and whether the formulation yielded single phase. Fully reacted samples were + Present

address: Departments of Chemistry and Materials

Science and Engineering, Ohio State University, Columbus, OH 43210-1173.

0897-4756/90/2802-0421$02.50/0

Table I. Phase Distributiona after 860 O C for 100 h in 0% batching compn cation

Bi l

A B

2 2 2 2 2 2

C D

E F G

content Sr Ca l 1 1.5 2 1.2 1.6 2

1 2 1.5 1 2 2 2

Cu 2 2 2 2 3 3 3

major phase A A A A A A A

minor phase U Ca2Cu03U

U CuO, CazCuO3, U U U

-

"A: pseudotetragonal Aurivillius phase a l b -5.4 A, c -30.6 A, 4.7. U: unknown(s) with strong lines at d = 3.41, 3.17, 2.96, 2.77, 2.46, and 2.35 A.

b*

annealed for 4 h at 500 "C in nitrogen, 0.2% oxygen, 5% oxygen, air, or oxygen and furnace cooled. Afterwards, T,was determined by ac susceptibility. The Bi2Sr1,5Ca1,5C~208 composition was characterized with respect to thermal expansion in oxygen and nitrogen as well as the compositional expansion while the atmosphere was switched during isothermal heating. Thermogravimetric (TGA) and evolved gas (EGA) analyses on this same composition completed the analyses. Lattice constants for different oxygen stoichiometries were determined from room-temperature and high-temperature XRD measurements. Results a n d Discussion P h a s e Distribution. Table I shows the phase distributions, obtained for the various compositions sintered at 860 "C in oxygen for 100 h. Sample melting and/or reaction with the Ag sheet became appreciable at 870 "C in air or 880 "C in oxygen for most compositions. The choice of a 100-h sintering time was made because shorter times often did not achieve full reaction as indicated by the changes in the relative phase amounts with time. XRD did not find second phase in composition C, and its indexed pattern is shown in Figure 1. The indexing includes a fourth Miller index, which is used to identify the lines that originate from an incommensurate m o d u l a t i ~ n . ~ * ~ This structural irregularity occurs along the b axis and is caused by the insertion of an extra oxygen atom in the Bi layer once every -4.8 unit cells.1° The lines are generally (1)Maeda, H.;et al. Jpn. J . Appl. Phys. 1988, 27, L209. (2) Endo, U.;et al. Jpn. J . Appl. Phys. 1988,27, L1476. ( 3 ) Takano, M.; et al. Jpn. J. Appl. Phys. 1988,27, L1041. ( 4 ) Cava, R. J.; et al. Phys. Lett. 1988, 153, 560. (5) Buckley, R. G.;et al. Physica C 1988, 156, 629. (6) Morris, D.E.;et al. Phys. Reo. B 1989, 39, 6612. (7) Groen, W.A.; deleeuw, D. M. Physica C 1989, 159, 417. (8) O n d a , M.; et al. Jpn. J . Appl. Phys. 1988, 27, L833. (9) Shaw, T. M.; et al. Phys. Reo. B 1988, 37, 9856.

0 1990 American Chemical Society

O'Bryan et al.

Chem. Mater., Vol. 2, No. 4, 1990

422

I .OOr

:

m

8-

0.00 m

W

2x

0.601 Lo

6-

0

2

-

1

n

0.20

-

4-

2-

0

0

20

10

30

50

40

60

0

d

l

1

I

1

I

I

I

I

I

I

1

1

I

I

I

280

Figure 1. XRD diffraction pattem for Bi~rl5CalbCu2Ow heated 860 O C for 100 h in O2and slow cooled. Indexing follows 0noda.B

1

I

I

I

1

20

40

60

00

400

4

"0

OXYGEN

Y

-

T(K)

4

"

0

200

400

600

800

O C

Figure 4. CTE of BizSr1,5Ca1,5Cuz0,. Heating rate 4 O C / h (a) oxygen; (b) nitrogen.

-5

-4

-3

-2

0

-1

log P o p

Figure 2. Effect of annealing atmosphere on T,(onset): (A) Bi&l,5Ca15Cu208annealed at 500 O C ; (B) T,(onset) as a function of Pol. Sample designation from Table I.

-

weak [the strongest being (0211) a t I / I 8% and 20 N 37.0°] and are often misidentified as second phase or ignored. For all compositions the major phase could be indexed with a pseudotetragonal unit cell having the average dimensions a = 5.4,b = 5.4,and c = 30.7 A and the b# 4.7 modulation factor. The identification of singlephase composition as Bi2Sr1.5Ca1.5CU308has also been reported by Tarascon et al." and Roth et a1.I2 Critical Temperature. As prepared, the T,(onset) was 70-78 K for the various compositions. After equilibration at 500 "C for 4 h in atmospheres with P%'s between 1atm and atm (nitrogen), susceptibility curves similar to Figure 2a were obtained. Figure 2b summarizes the onset Tc's for several compositions. Quenching or longer an-

-

(10) Lapage, y.;et al. Phys. Rev. E 1989,40,6810. (11) Tarascon, J. M.; Phys. Reu. E 1988,37,9382. (12) Roth, R. S.;et al. Paper 6 SIII-90 92th Annual Meeting, American Ceramic Society Dallas, April 23, 1990. (13) O'Bryan, H.M.; Gallagher, P. K. MRS Symposium M, 1989.

nealing times did not affect the T,. However a strong effect of Po2on T, is noted, and low Po,raises T,. XRD measurement on these samples did not disclose that annealing produced any phase change or alteration of the incommensurate lines, but a small increase (+0.1 A) was obtained in the c axis for nitrogen vs oxygen. The fact that for similar heat treatments both the two-phase and singlephase samples have the same T, (onset) suggests that either the superconducting phase has no solid solution region or there is no effect of cation stoichiometry on T,. Thermal Analyses. Dilatometric data for a ceramic sample with t h e single-phase composition, Bi2Srl.5Ca1.5C~308+6, showed that its expansion depends strongly on the oxygen partial pressure in the atmosphere. For a sample that had been furnace cooled from 860 "C after a 100 h in oxygen, the initial expansion curve (Figure 3) shows an increased rate of expansion near 200 "C. Reheating oxidized samples in either oxygen or nitrogen did not reproduce this low-temperature anomaly, but changes in slope near 350 "C were seen. An increase in the coefficient of thermal expansion (CTE) from 9 x lo4 to -14 X 10+ ("C)-l near 350 "C is shown in Figure 4a. The more pronounced change in CTE during heating suggests an oxygen stoichiometry change that occurs quickly with increasing temperatures and slowly with falling temperature. When a sample cooled in oxygen is reheated in nitrogen, the thermal expansion again increases near 325 "C, and the CTE reaches 18 X lo4 ("C)-l before falling to 14 x lo* ("C)-l (Figure 4b). On cooling in ni-

Two-Layer BiSrCaCu Oxide Superconductors

Chem. Mater., Vol. 2, No. 4, 1990 423

0 00

:-0.04

I

7c

0.04

1

I

--- 360'C

6

--O5OC

-0.00I

1

I

I

/HEATING

f

0

~

0

1

1

1 1I 1

1

400

1

1

1

l

l

200

I

I

1

I

I

l

l

I

l

400

300

1

I

500

TIME ( m i n )

Figure 7. Expansions of BizSr1,5Ca1.5C~ZOx at 505 and 360 O C

as atmosphere is switched.

"

0.1

0.05

0

0.03

-

r

2

z 0.oi 0 I

1200

I

1

1300

I

1400

1

I

1500

TIME ( m i n )

Figure 6. Change in oxygen stoichiometry of Bi2Sr1.5Ca1.SC~20r as atmosphere is switched at 500 O C .

0 O C

trogen, no change in expansion rate near 350 "C is observed because oxygen is not available for recovery. TGA measurements on a powder obtained by crushing a sintered sample are shown in Figure 5. During heating weight loss begins for both atmospheres near 350 and 500 "C the oxygen content has decreased by A6 = 0.02 in oxygen and A6 = 0.06 in nitrogen. Upon cooling the lost oxygen is regained only in the oxygen atmosphere. An anomalous effect was noted for a sample been exposed to the laboratory atmosphere. For this sample the initial heating produced a weight loss of 0.3% or A6 = 0.16 below 400 "C, and EGA has identified the evolved material as water. Subsequent heating and cooling in oxygen or nitrogen did not show this weight change. This water evolution must also be the source of the anomalous expansion noted in dilatometry for the initial heating (Figure 3, curve 1). Isothermal TGA and dilatometry confirm that changes in oxygen stoichiometry produce the expansion differences noted. Figure 6 shows that at 500 "C a sample equilibrated in nitrogen gains weight rapidly when the atmosphere is switched to oxygen. The gain of 0.075 wt % corresponds to a change in oxygen stoichiometry A6 = +0.04. When the atmosphere is switched back to nitrogen the weight loss is also 0.075% (A6 = -0.04) but an hour is required to reach a stable weight. At 600 "C, A6 = 0.043 and the kinetics are faster. Isothermal dilatometry (Figure 7) shows that the addition of oxygen produces linear shrinkages of 0.65 X and 0.40 X at 505 and 360 OC, respectively. These shrinkages are reversed when the atmosphere is switched to nitrogen. The kinetics of oxidation are considerably faster than those for reduction in agreement with the TGA results. Table I1 shows the lattice constants and expansions obtained by XRD at 25 and 500 "C for the different oxygen contents. The dilatometric expansions (D)are included

Figure 8. Expansion of Bi2Srl.5Ca1.SC~20r during heating in oxygen to 500 "C, equilibration with nitrogen at 500 "C for 4 h, and cooling in nitrogen.

Table 11. Lattice Constants and Expansions for Samples Annealed for 4 h at 500 OC Lattice Constant oxygen nitrogen XRD temp, "C u C bt U C b# 25 5.4059 30.668 4.67 f 0.1 5.4113 30.773 4.74 f 0.1 500 5.423 30.928 5.425 30.963

Thermal Expansion (%), 25 av AL/L(XRD) AIL (D)

0.63 0.58

-

compositional expansion, Oxygen av AL/L (XRD) 25 O C 500 O C AL/L(D) 500 O C

5OOOC

0.53% 0.53

Nitrogen 0.18% 0.06% 0.065%

for comparison. Agreement between the thermal expansions, measured by XRD and dilatometry, is good. Also in good agreement are compositional expansions for measurements made at 500 "C. The larger compositional expansion calculated from the 25 "C constants is partially explained by the increased thermal expansion of the oxygen-loaded Bi2Srl,5Ca~,5C~20w6. This is shown graphically in Figure 8, where the length difference between nitrogen and oxygen is seen to be 50% larger a t 25 "C that at 500 "C. At temperatures 2700 "C, the TGA data show a slow weight loss during a nitrogen anneal (Figure 9). Upon a switch to oxygen, there is a rapid recovery of oxygen followed by a slower recovery. The weight changes are fully reversible. Dilatometry at 700 "C shows a slow shrinkage in both oxygen and nitrogen that is not recoverable. The length change is caused by either a deformation of the

O'Bryan et al.

424 Chem. Mater., Vol. 2, No. 4, 1990

00

K

750°C

co Q

-0.2

-0.4

TIME ( m i n )

Figure 9. TGA of BiPSrl,6Ca1,5C~POI as atm is switched a t 750 "C.

Table 111. Effect of Oxygen Stoichiometry this study single-phase composition Po2 range

T,ac susceptibility, K c axis, A Ac axis

A6 isothermal U / L expansion 25 to 600 "C, 70

BiZSr1,5Ca1.5-

Groen7 Bi~SrzC~.gYo.iC~z0~+*

CU208+b to 1 atm at to atm at 600 OC 500 OC 90 to 66 (onset) 88 to 61 (midpoint) 30.792 to 30.668 -0.12 t0.040 -0.065 0.66 (PO, = 1 atm)

30.890 to 30.823 -0.07 t0.04 (Pot 10-3.7to 1 atm) -0.06 0.21 (po, = lo4,' atm)

sample under the weight of the dilatometer probe or a reaction between the silica holder and the cuprate and is not related to the slow weight change observed by TGA. A t lower temperature all expansion changes were reversible. The weight change noted by TGA appears to be caused by a phase decomposition, although no second phase appears in the XRD pattern. It is suggested that 750 "C is too low a temperature for the formation of the second-phase regions of the size necessary to produce an XRD pattern. In summarizing the results of this study, it is instructive to compare our data with those of Groen. Although the

compositions that yield a single-phase two-layer ceramic are different, the effects of annealing in oxygen and nitrogen are quite similar. Table I11 compares those results. With regard to the thermal expansion, the value obtained by Groen for air seems too low and probably has been reduced by a small peak that he observed near 300 "C during initial heating and attributed to a phase transition. It is probable that this anomaly was caused by the evolution of moisture as was observed in our study. We note a substantial difference in thermal expansion for different 8 but observe similar compositional expansion. It has been suggested that in the 2212 structure the extra oxygen (i.e., 6) required for superconductivity is accommodated by the insertion of oxygen in the Bi layers. The result is a corrugation of these layers.1° For a incommensurate factor b# 4.7 there is -0.2 atom extra per unit cell. If the oxygen removed during annealing is taken from the Bi layers, then a change in b# is expected. The fact that the present data (Table 11) show that the incommensurate factor is not affected by the A6 = 0.04 suggests another source. However, since it is difficult to measure b# and only 20% of the inserted oxygen has been removed, the data are not conclusive. In both the Ba2YC~307-6 and Pb2SrzYCu308+6compounds a change in oxygen stoichiometry is seen to affect the c-axis expansion.12 In both structures the removable oxygen is located near Cu, which has 2-fold or 4-fold coordination. However only in Ba2YC~307-6 does a lower oxygen content increase the lattice constant and lower T,. Since all the copper ions in the Aurivillius structure (2212) have 5-fold coordination, it is difficult to find the removable oxygen in a position where Cu coordination can be increased. The relatively larger effect of composition on expansion for Bi2Sr1.5Ca1.5C~308+6 vs Ba2YC~307-6 (AL/ LA8 = 1.5% /atom vs 0.35%/atom) also suggests a site not closely connected to copper, and perhaps the removable oxygen is positioned between the puckered Bi layers.

-

Acknowledgment. Discussion with S. A. Sunshine on the role of structure is greatfully acknowledged, and R. M. Fleming and P. Marsh are thanked for computing assistance.